Correlated Chained Gaussian Processes for Datasets With Multiple Annotators.

The labeling process within a supervised learning task is usually carried out by an expert, which provides the ground truth (gold standard) for each sample. However, in many real-world applications, we typically have access to annotations provided by crowds holding different and unknown expertise levels. Learning from crowds (LFC) intends to configure machine learning paradigms in the presence of multilabelers, residing on two key assumptions: the labeler's performance does not depend on the input space, and independence among the annotators is imposed. Here, we propose the correlated chained Gaussian processes from the multiple annotators (CCGPMA) approach, which models each annotator's performance as a function of the input space and exploits the correlations among experts. Experimental results associated with classification and regression tasks show that our CCGPMA performs better modeling of the labelers' behavior, indicating that it consistently outperforms other state-of-the-art LFC approaches.

CNN-based framework using spatial dropping for enhanced interpretation of neural activity in motor imagery classification.

Interpretation of brain activity responses using motor imagery (MI) paradigms is vital for medical diagnosis and monitoring. Assessed by machine learning techniques, identification of imagined actions is hindered by substantial intra- and inter-subject variability. Here, we develop an architecture of Convolutional Neural Networks (CNN) with an enhanced interpretation of the spatial brain neural patterns that mainly contribute to the classification of MI tasks. Two methods of 2D-feature extraction from EEG data are contrasted: Power Spectral Density and Continuous Wavelet Transform. For preserving the spatial interpretation of extracting EEG patterns, we project the multi-channel data using a topographic interpolation. Besides, we include a spatial dropping algorithm to remove the learned weights that reflect the localities not engaged with the elicited brain response. We evaluate two labeled scenarios of MI tasks: bi-class and three-class. Obtained results in an MI database show that the thresholding strategy combined with Continuous Wavelet Transform improves the accuracy and enhances the interpretability of CNN architecture, showing that the highest contribution clusters over the sensorimotor cortex with a differentiated behavior of rhythms [Formula: see text] and [Formula: see text].

Feature relevance analysis supporting automatic motor imagery discrimination in EEG based BCI systems.

Recently, there have been many efforts to develop Brain Computer Interface (BCI) systems, allowing identifying and discriminating brain activity, as well as, support the control of external devices, and to understand cognitive behaviors. In this work, a feature relevance analysis approach based on an eigen decomposition method is proposed to support automatic Motor Imagery (MI) discrimination in electroencephalography signals for BCI systems. We select a set of features representing the best as possible the studied process. For such purpose, a variability study is performed based on traditional Principal Component Analysis. EEG signals modelling is carried out by feature estimation of three frequency-based and one time-based. Our approach provides testing over a well-known MI dataset. Attained results show that presented algorithm can be used as tool to support discrimination of MI brain activity, obtaining acceptable results in comparison to state of the art approaches.

Training using short-time features for OSA discrimination.

Heart rate variability (HRV) is one of the promising directions for a simple and noninvasive way for obstructive sleep apnea syndrome detection (OSA). The interaction between the sympathetic and parasympathetic systems on the HRV recordings, gives rise to several non-stationary components added to the signal. Aiming to improve the classifier accuracy for obstructive sleep apnoea detection, the use of more appropriated techniques for leading with non-stationarity and mixed dynamics, are needed. This work aims at searching a convenient training strategy of combining the feature set to be further fed in to the classifier, which should take into consideration the different dynamics in the HRV signal. Therefore, a set of the short-time features, extracted from a given HRV time-varying decomposition, and selected by spectral splitting is considered. Additionally, three methods of projection are used: none, simple, and multivariate. Finally, the different approaches are tested and compared, using k-nn and support vector machines (SVM) classifiers. Attained results show that using continuous wavelet transform with short-time features and multivariate projection, followed by a SVM classifier, allow to obtain a suitable OSA detection.

Biomedical data analysis by supervised manifold learning.

Biomedical data analysis is usually carried out by assuming that the information structure embedded into the biomedical recordings is linear, but that statement actually does not corresponds to the real behavior of the extracted features. In order to improve the accuracy of an automatic system to diagnostic support, and to reduce the computational complexity of the employed classifiers, we propose a nonlinear dimensionality reduction methodology based on manifold learning with multiple kernel representations, which learns the underlying data structure of biomedical information. Moreover, our approach can be used as a tool that allows the specialist to do a visual analysis and interpretation about the studied variables describing the health condition. Obtained results show how our approach maps the original high dimensional features into an embedding space where simple and straightforward classification strategies achieve a suitable system performance.